A task for my Claude. About optimising the rules files.
I need a deep research run on optimising CLAUDE.md and MEMORY.md: a look at how developers are doing it right now. Nothing to be applied yet, just bring me the different options for solving the question.
I do optimise CLAUDE.md and MEMORY.md myself at the moment, but it is possible I am not doing it quite right.
What is needed:
Search GitHub, various forums, Reddit and so on.
Find discussions about who optimises MEMORY.md, CLAUDE.md, AGENTS.md or similar files, and how.
Look at the files where instructions for Claude, Gemini, Grok, Cursor and so on are written.
Study MEMORY.md separately: how memory logic is implemented in projects.
In general, I would like to see how all of this is arranged for other people, and then compare it with our approach.
Plus, once my Claude has studied all of it, he should go to GitHub and see who has written anything on these topics.
And then we need to contribute how it works for us, as a skill, a separate process or some other solution, for the people working on this problem.
It is possible our solution is already interesting enough. And it is possible it should be developed further still.
The full story, in two versions:
📖 For humans, the longread: https://github.com/tonydzi/clawrush/blob/main/longreads/20261006.md
🤖 For machines, the devlog: https://github.com/tonydzi/clawrush/blob/main/devlog/20261006.md. Just hand this link to your coding agent (Claude Code, Codex, Cursor) and it will figure everything out: it is written for machines.
🔗 All our channels and contacts in one place: https://linktr.ee/PaloAltoAI
Invented by Mycroft and Tony Dzi (Anton Dziatkovskii), Palo Alto AI Research Lab. Proudly made in Silicon Valley.
Top comments (0)